Re:Numbers changed drastically (Score 1)

I've been tracking the data on this asteroid for a few days now. 26 DEC 2004 - Based on 161 Observations they determined that the probability was 1 in 45. Early 27 DEC 2004 - Based on 176 Observations they determined that the probability was 1 in 37. And now Based on 118 Observations they determined that the probability is 1 in 24,000. I'm just curious as to why discard any observations at all? How is one observation more accurate than the next? Is there any source for acquiring all observational material? Was that data just discarded simply because new obveservational data was capable of pinpointing errors within past data thus identifying it as 'flawed', or 'inaccurate'? And in response, if they can just discard inaccurate observations then can't they just discard accurate ones as well? Already we saw how many sites and newspapers were carrying this story, and it had only been rediscovered for analysis recently. There was discussion about how the JPL and the B612 Foundation's proposed planetary defence missions could be accelerated in the event of the MN4's continuing threat. I would say that I am not completely satisfied with how quickly these stats turned from 1 in 35 to 1 in 25,000 in a matter of hours, using less observations, and without them providing any real explanation as to why those observations were discarded.